A total of 42 female endurance athletes (23 medium- and long-distance runners and 19 cross-country skiers) and 45 non-athlete light physically active female university students between the ages of 15–30 consecutively examined in the Sports Research Institute of the Czech Armed Forces between 2021 and 2022 were included in the study. Athletes and control women who used hormonal contraceptives, pregnant or lactating women were excluded from the study. Based on the evaluation of the LEAF-Q data and the interest of the subjects examined in the further participation in the study, the individuals were divided into two study groups: endurance athletes at risk of LEA (LEAF-Q score ≥8) and control participants not at risk of LEA (LEAF-Q score <8).

All athletes with LEAF-Q ≥ 8 were interested in additional participation in the study, therefore, the athlete study group with LEAF-Q score ≥8 consisted of 23 women athletes (16–29 years, Table 1), 16 runners, and 7 cross-country skiers. The subjects were representatives of endurance sport disciplines mainly from youth sports centers and all were of Caucasian origin.

Not all women in the control group who had LEAF-Q < 8 were interested in further participation in the study (14 controls with LEAF-Q < 8 and 10 controls with LEAF-Q ≥ 8 were therefore excluded from further examinations). Therefore, the controls consisted of 21 healthy women with a LEAF-Q score <8 of Caucasian origin (20–30 years, Table 1). The women were lightly physically active (1–3 days a week), and none of the control women had a sports background. These two groups, selected according to the LEAF-Q score, were further examined in detail.

The study was designed to intentionally compare endurance athletes at risk of LEA with lightly physically active control participants without LEA. This approach was chosen to minimize the heterogeneity related to training load and to allow comparison with a physiologically representative reference group.

The whole study was carried out in accordance with the Declaration of Helsinki of 1975, as revised in 1983. The athlete study was approved by the CASRI Ethics Committee (No. 6/1–6/8 2019). All participants signed an informed consent. Before recruitment, all participants received detailed information about study objectives, which was also included in the informed consent form. Participants under 18 years of age were included in this study with parental consent.

LEAF-Q in study participants was used as previously described (9). This standardized questionnaire aims to address symptoms of insufficient energy intake and is used for screening female athletes to identify individuals at risk of the female athlete triad. The cut-off value of LEAF-Q of 8 discriminating those at risk of the female athlete triad was also used in our study (9). The questions of LEAF-Q are divided into three areas that cover injury, gastrointestinal, and reproductive functions. The questionnaire was translated into Czech, uploaded to an online survey platform (Microsoft Forms, Microsoft Corporation, USA), and evaluated and scored as described (9). The LEAF-Q data from the online questionnaire were collected from all participants as described above.

Body composition (weight, BMI, body fat, muscle mass, and FFM) was evaluated using bioelectrical impedance analysis technology on an automatic body composition analyzer (Tanita MC-780, Tanita Corp., Japan). All measurements were made by trained personnel using a standardized technique. The resting metabolic rate (RMR) was not measured directly but was estimated using the Katch & McArdle formula (10). The total daily energy expenditure was also estimated by multiplying RMR by a standard activity factor. We used a uniform factor for athletes of 1.725 (hard exercise/sports 6–7 days a week) and non-athletes of 1.375 (light exercise/sports 1–3 days a week) (11). The selected activity factors were justified based on the characteristics of the training program, including intensity, type, and frequency of exercise. The factor for non-athletes reflect light to moderate physical activity levels typical of recreationally active individuals without a sporting background. Exercise energy expenditure was calculated as the energy expenditure per day minus RMR. EA reflects the difference in energy intake and exercise energy expenditure in relation to FFM.

Participants enrolled according to LEAF-Q scores completed dietary records using the NutriData.cz web application (Fitsport-komplex s.r.o, Czech Republic). Dietary intake was recorded for a minimum of four complete days, including at least one weekend day, within a 14-day period in order to capture habitual intake and potential variability. Participants were instructed to record all foods and beverages consumed, including portion sizes. Portion sizes were estimated using household measures or standardized portion size guides provided within the application. Dietary records were subsequently analyzed using NutriPro nutritional software (Fitsport-komplex s.r.o, Czech Republic). Total energy intake and macronutrient intake were calculated for each participant and averaged across recorded days. Nutrient intake was evaluated in relation to dietary reference values as previously described (12, 13).

Venous blood samples were collected in the morning after fasting overnight and analyzed in the laboratory within 1 h after blood sampling. An aliquot of serum was immediately stored at −80 °C for later hormonal analyses. The determination of serum biochemical and hematological parameters was performed on automatic analyzers (Chemistry Analyzer BS-240 and Hematology Analyzer BC-3600, Mindray Bio-Medical Electronics, China).

All hormonal parameters, with the exception of 25-hydroxyvitamin D and leptin (see below), were determined on an automatic analyzer using enzyme immunoassays according to the manufacturer's instructions (DRG:HYBRiD.XL®, DRG International, Inc., United States).

Serum leptin concentrations were determined using the enzyme immunoassay DRG Leptin Sandwich ELISA (EIA-2395, DRG International, Inc., United States), absorbance was detected by the Infinite F50 Plus ELISA reader (Tecan Group Ltd., Switzerland), and concentrations were calculated using software Magellan™ (Tecan Group Ltd., Switzerland).

The concentration of 25-hydroxyvitamin D was determined using a direct particle-enhanced immunoturbidimetric method on an automatic analyzer (Chemistry Analyzer BS-240, Mindray Bio-Medical Electronics, China) using diagnostic reagents according to the manufacturer's instructions (Dialab spol. s.r.o., Czech Republic).

The phase of the menstrual cycle was determined according to the following criteria: subjective evaluation of the phase of the cycle by the research participants and the concentration of luteinizing hormone (LH) helped reveal whether the woman is currently ovulating. Women with LH values greater than 15 mIU/mL (reference cut-off value for serum LH concentrations that discriminate between the follicular and luteal phase) and regular menstruation were reassigned to the ovulatory phase group. Women with LH values lower than 7 mIU/mL and regular menstruation were reassigned from the ovulatory phase group to the luteal phase (lower follicle-stimulating hormone, FSH) or the follicular phase (higher FSH) group according to the FSH value. In eumenorrheic women, blood sampling for determination of sex hormones was timed according to the phase of the menstrual cycle determined as described above, while in amenorrheic women blood sampling was performed randomly.

Two female athletic runners were supplemented with levothyroxine due to hypothyroidism. One cross-country skier, nine athletic runners, and two women in the control group were taking iron supplementation (commonly available dietary supplements in an average dose of 20 mg per day). Two cross-country skiers, nine athletic runners, and a woman in the control group were taking vitamin D supplementation (generally available dietary supplements from 1,000 IU to 2,000 IU per day). For the 25-hydroxyvitamin D, hepcidin, and iron metabolism parameters, only those who did not take vitamin D and iron supplements were compared. The study participants did not take other supplements or acute or chronic medication.

The Shapiro–Wilk normality test and descriptive statistics were used to evaluate the distribution of the data. Most laboratory variables did not follow normal distribution. Therefore, to ensure coherence and avoid ambiguity, all laboratory data are presented as median and interquartile range. Normally distributed data on study group characteristics (Table 1) are presented as mean ± standard deviation. Categorical data are expressed as a percentage of the specific group. The T-test or the Mann Whitney U-test was used to compare laboratory parameters. The frequency of the LEAF-Q score ≥8 or <8, absence from training/races due to injury, age of the first menstruation, frequency of menstrual cycles during the last year between athletes and a control group, and frequencies of individual genes variants between athletes and a control group were determined using Pearson's chi-square test (χ²). All analyses were performed with the alpha set to 0.05. Statistical analyses were performed using Statistica Ultimate Academic v. 14 (TIBCO Software Inc. California, USA).

Compared to the control group (n = 45), a higher LEAF-Q score was observed in athletes (n = 42) (8 vs. 5, p < 0.005, Table 2), with a much higher percentage of LEAF-Q score ≥8 in athletes (54.8% vs. 22.2, p < 0.005). Athletes also missed training sessions during the last year due to musculoskeletal problems much more common compared to the control group (57.1% vs. 33.3%, p < 0.05). Similarly, menstrual problems were significantly more frequently reported in athletes; 33.3% of the athletes had amenorrhea, others had oligomenorrhea, and only 31% of the athletes had a normal menstrual cycle (i.e., 12 or more cycles/year) during the year (Table 2). Athletes menstruated later than the general population; 28.6% of athletes were at the age of 15 years or older.

As described above, according to the questionnaire data, the subjects were divided into a group of female athletes with a LEAF-Q score ≥8 (n = 23) and a group of control women who had a LEAF-Q score <8 (n = 21).

The anthropometric characteristics of the studied groups are given in Table 1. Compared to the control group with LEAF-Q < 8, athletes with the LEAF-Q score ≥8 had lower body fat (19.9 vs. 22.4%, p < 0.005), BMI (19.7 vs. 22.2 kg/m², p < 0.001) and weight (56.8 vs. 61.5 kg, p < 0.05). The athletes were taller (170 vs. 166 cm, p < 0.05) and spent more time training (12 vs. 5 h/week, p < 0.001).

Compared to control women with LEAF-Q < 8, total and exercise energy expenditure was higher (p < 0.001) in athletes with LEAF-Q ≥ 8 and athletes had a negative energy balance (90%) with a low EA value (24,3 kcal/kg FFM/day) (Table 3). Surprisingly, the control women also, although with a low LEAF-Q score and steady energy balance, had a borderline value of EA (29.8 kcal/kg FFM/day). Interestingly, the dietary intake of vitamin D, iron, and folate was below the recommended daily intake (RDI) in both groups (Table 3). However, in a group of athletes, there was a higher dietary intake of iron (p < 0.05) compared to the control group and the athletes paid more attention to iron and vitamin D supplementation than a control group (10 athletes vs. 2 women in the control group took iron supplementation, 11 athletes vs. 1 woman in the control group took vitamin D supplementation). Compared to controls, athletes had a higher consumption of carbohydrates (p < 0.05), and proteins (p < 0.001) and a lower intake of fats (p < 0.001) (Table 3). However, the athletes still had an inadequate recommended dietary allowance (RDA) for carbohydrate intake (Table 3).

The blood serum biomarkers are given in Table 4. As described above for the 25 hydroxyvitamin D, hepcidin, and iron metabolism parameters, only those who did not supplement were compared. Concentrations of 25 hydroxyvitamin D were at the lower end of the reference range and comparable between both groups. Compared to the control group with LEAF-Q score <8, athletes with LEAF-Q score ≥8 had lower serum concentrations of leptin (p < 0.001), white blood cells (p < 0.005) and phosphorus (p < 0.005), and higher concentrations of hepcidin (p < 0.05) and free T3 (p < 0.05). The concentrations of free T4 and TSH did not differ and were within the reference limits. Hypophosphatemia (serum phosphorus concentration ≤0.8 mmol/L) occurred only in 3 athletes. The fasting serum leptin concentration was lower in the athlete group (p < 0.001) and was positively correlated with adiposity (percentage of body fat) (r = 0.65, p < 0.001) and negatively correlated with time spent on training/exercising (r = −0.5, p < 0.05) in all subjects.

Athletes also had lower estradiol and progesterone concentrations compared to the control group (p < 0.001) (Table 4). Almost 50% of the athletes had an irregular or absent menstrual cycle. In accord, the estradiol concentration in athletes generally moved around the lower reference range in individual phases of the cycle (Table 5). There were lower concentrations of progesterone (p < 0.05) in the luteal phase in athletes compared to the control group, and this concentration was below the lower limit of the reference range (Table 5). When compared eumenorrheic and amenorrhoeic/oligomenorrheic subjects, significantly different concentrations of estradiol (239 pmol/L vs. 117 pmol/L, p < 0.005) and progesterone (2.68 nmol/L vs. 1.41 nmol/L, p < 0.005) were observed (Table 6).